China Suddenly Follows Trump's Lead on AI Rules, Global Impact Expected
The Chinese government is reportedly drafting new regulations that could restrict foreign access to the most advanced artificial intelligence models developed by domestic companies, including those not yet officially launched. This protectionist move follows a similar, though later rescinded, action by the Trump administration, which had asked Anthropic to bar foreign nationals from accessing its most sophisticated models.
According to Reuters, Chinese authorities have held a series of meetings with major technology firms over the past month to discuss the policy. The initiative is part of Beijing’s effort to tighten protection of AI technology, which is now viewed as a strategic national asset, mirroring the approach taken by the United States. Companies involved in the discussions include Alibaba, ByteDance, and AI startup Z.ai. Sources familiar with the talks indicated they are still in the early stages and no final decision has been made.
The policy push comes as Chinese AI models, particularly since the emergence of DeepSeek R1 last year, have gained significant global traction. Their lower costs and improving capabilities have made them competitive with US products. If Beijing restricts foreign access to these models, the global AI market could face disruption, with many companies potentially facing higher costs due to the loss of competitively priced Chinese alternatives.
During the meetings led by the Ministry of Commerce, participants discussed potential restrictions on the most advanced AI models, both closed-source and more open ones. Officials also discussed the possibility of making the leakage or theft of AI technology a criminal offence under China’s strict national security laws. The government is also considering new rules to limit which parties can fund domestic AI startups. The scope of the regulation is still under discussion and may only apply to future generations of AI models. There is no certainty on when or if the rules will be implemented.
The Ministry of Commerce and the National Development and Reform Commission (NDRC) have not responded to requests for comment, nor have Alibaba, ByteDance, or Z.ai. These companies possess various AI models, including closed-source and open-weight systems that allow users to download, run, and modify the base systems. Alibaba’s Qwen and ByteDance’s Doubao are currently the most widely used AI models in China, while Z.ai’s GLM-5.2 has recently attracted attention in Silicon Valley for its near-leading capabilities at a much lower cost.
Two sources indicated that the Chinese government is particularly concerned about the capabilities of Anthropic’s Mythos model to exploit software security vulnerabilities. Beijing is reportedly worried that Washington could use such models to attack Chinese interests. This concern aligns with statements from state media and Zhou Hongyi, founder of cybersecurity firm 360, who has said China must develop its own version of Mythos to avoid falling behind in the AI race.
Throughout the year, Beijing has been strengthening protections for its domestic AI industry. In April, the state planning agency ordered Meta to cancel a $2 billion acquisition of AI startup Manus, which was founded by a Chinese citizen. In early June, the government issued new rules tightening oversight of foreign transactions involving investors, technology, data, and national security interests. Authorities also launched an investigation into Manus and other local AI startups that moved their operations overseas to determine if they violated export control regulations.